BBC Royal Institution Christmas Lectures - The Truth about AI (2023)

Professor Mike Wooldridge asks: what is artificial intelligence? He compares how AI works and learns with how the human brain functions. Exploring the roots of AI, Mike reveals how Alan Turing devised the Imitation Game – a test of whether a machine answering a series of questions could pass as a human. The audience in the lecture theatre play a real-life version of the game to find out if AI can pass this test today.

Part 1: How to Build an Intelligent Machine
Mike Wooldridge examines real-life neurons in action and explains how artificial neural networks are inspired by neural structures in the brain. To demonstrate how AI learns, we watch drones as they are trained to recognise and fly through structures in the lecture theatre autonomously. AI exploded into the public consciousness in 2022 with the release of ChatGPT and boasts around 100 million monthly users. Mike unravels the mystery of how large language models like ChatGPT work, and he finds out if one day this technology - along with a whole suite of different AI tools - will allow us to understand the animals we share this planet with.
The Christmas Lectures are the most prestigious event in the Royal Institution calendar, dating from 1825, when Michael Faraday founded the series. They are the world's longest running science television series and always promise to inspire and amaze each year through explosive demonstrations and interactive experiments with the live theatre audience.

Part 2: My AI Life
Mike Wooldridge reveals the huge role AI already plays in our daily lives – sometimes without us even realising its role. Mike investigates how games like chess and Go have become a training ground for AI, helping to bring about key advances we are now seeing in the field, and he reveals how simple methods of learning, like rewarding success, have been used to train AI in spectacular ways. We also feature some of the revolutionary innovations that AI has brought about in healthcare, from the use of AI tools in planning cancer treatment, to monitoring Parkinson's. Mike is joined by members of DeepMind's AlphaFold team, who use AI to predict the structures of large numbers of proteins, which will revolutionise the creation of new drugs across the world.
We also reveal the huge impact AI has had on our creative lives – as it is able to write songs and create artworks in seconds. With the help of artist Eric Drass (aka shardcore), the audience creates a collaborative artwork and discovers how image generation works. Mike explores the thorny question of who the creator is – the AI itself, the human who set it to work, or the creators of the art that AI has learned from? The Christmas Lectures are the most prestigious event in the Royal Institution calendar, dating from 1825, when Michael Faraday founded the series. They are the world's longest running science television series, and always promise to inspire and amaze each year through explosive demonstrations and interactive experiments with the live theatre audience.

Part 3: The Future of AI: Dream or a Nightmare
Mike Wooldridge grapples with the future of AI in the third and final Christmas lecture. Mike takes a ride in a driverless car. Autonomous vehicles, once a science fiction dream, are now a reality. Many AI researchers believe removing human drivers will eventually make our streets much safer. Mike explores how the car ‘sees' and perceives the world – and how with the help of AI, it gets better the more it drives. Although AI will create many exciting opportunities, advances in AI have raised fears – some justified, others not. With the help of expert guests, Mike talks us through some of the risks AI poses. He unpacks the very real danger of bias in AI, asking how we avoid creating AI that favours those who resemble its creators, and he explores the dangers of 'fake news' and how AI algorithms can lead to dangerous online 'echo chambers', helping to foment extreme views. Mike also demonstrates deepfake technology and asks if AI means we simply can't trust our eyes any more? The prospect of super-intelligent AI means that in the future we may be able to mobilise AI to uncover radical large-scale solutions to the biggest problems facing humanity, such as climate change. But we need to think carefully about what we want to let AI control. Could AI in charge of weapons accidentally begin wars – and present a risk to our survival? And, as AI gets ever more intelligent, how should we treat it? How does our audience feel about kicking an AI robot dog? This lecture addresses the big question of AI: can it ever truly be like us, or are humans unique? As AI advances, it seems these ethical questions are destined to get ever more complex…